Application SysAdmin Guide - "How to reset the cluster without re-installation"

[norton]

Add new section to describe the steps for resetting a cluster without re-installation.

(2011/03/25 11:34), Joseph Wayne Norton wrote:

Miyasaka-san -

Try the following order of operations:

1. hibari/bin/hibari stop
2. Check by "ps -Alf | grep beam.smp"

If anything is running, "killall -9 beam.smp"

If you want to be really safe, "killall -9 epmd"
... but not recommended unless you know that hibari is the only Erlang
virtual machine running on the host.

1. Delete files hibari/data/brick/*
2. Delete files hibari/Schema.local
3. hibari/bin/hibari start
4. hibari/bin/hibari status

If status is NOT OK, then check logs, etc.

1. do bootstrap operations

NOTE: All brick nodes must be running before doing bootstrap operations.
NOTE: All brick nodes must have same cookie before doing bootstrap
operations.

If status is NOT OK, then check logs, etc.

If problem cannot be resolved, please do a FULL re-install and then
repeat the above procedure once and then repeat again to make sure.